Most LLM interactions are stateless. You explain your context, get a response, and start from scratch next time. Engram Wiki fixes that by storing your professional knowledge — projects, people, decisions, meeting history, writing style, and communication preferences — as markdown files in a git repository. When those files are available to an AI assistant, the model doesn’t just answer questions; it answers them as someone who already knows your context.
Git tracks changes to code. This pattern tracks changes to a person’s knowledge-work in a machine-readable format — every commit a snapshot of what you knew, decided, and were paying attention to.
Why “engram”?
An engram is the physical trace of a memory in the brain — the enduring network of neurons that changes when you experience something, letting you store and recall it later. The repo is the same idea for your work: a durable, machine-readable memory trace an AI can store and recall.
What it provides
The repository ships intentionally empty — structure, templates, and AI behavioral instructions only, no real content. It establishes a baseline directory model that adapts to any knowledge-work role:
- daily/ — append-only log of raw signal and observations
- people/, projects/, topics/ — living documents on collaborators, initiatives, and durable concepts
- decisions/ — ADR-style logs, immutable once written
- preferences/ and style/ — standing instructions for tone, defaults, and house style
- templates/, indexes/, exports/ — scaffolding for consistent, cross-referenced entries
The AI’s canonical instructions live in CLAUDE.md (read automatically by Claude Code), with a pointer for Copilot-based assistants.
Acknowledgements
The pattern is an independent reconstruction of ideas from Andrej Karpathy on LLMs as operating systems with persistent context, and Nate B Jones on structured personal knowledge management with AI copilots.
The project is open source under the MIT license. View the repository on GitHub.